• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文学位 >Stochastic techniques for the cooperative navigation of autonomous mobile robots.
【24h】

Stochastic techniques for the cooperative navigation of autonomous mobile robots.

机译：自主移动机器人协同导航的随机技术。

获取原文
获取原文并翻译 | 示例
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

The cooperative navigation system (CNS) algorithm is based on a Kalman filter which uses inter-robot relative-displacement measurements to improve the collective position estimates of a group of autonomous mobile robots. The principal motivation is reduction of the unbounded growth of position errors when dead-reckoning is used as the sole means of navigation for individual robots. In this thesis, dynamics-based models of groups of two-drivewheel robots are created and used in extensive simulations to verify the navigation improvement properties of the CNS algorithm. Observability properties of such groups of robots are also proven, and used to predict the scenarios in which the navigation errors remain bounded, as well as those wherein the errors are unbounded.; A simple quantitative relationship has been found for the lower bound of error reduction with a CNS group composed of mobile robots moving in formation. Simulations (of multiple robots) and experiments (with a pair of small two-drivewheel robots) reveal that this lower bound can be approached under the right conditions even with delays between inter-robot measurements.; A description of two hardware development projects essential for the experimental part of the research is also incorporated in the thesis. The first was the creation of an instrument platform for the rangefinder robot that allows it to obtain the correct range and bearing of the beacon-carrying reference robot, while ignoring other objects in the environment. The second hardware development task was the creation of a controller for the rangefinder robot consisting of two Texas Instruments (TI) digital signal processors (DSPs): a TI TMS320F243 and TMS320F2812 in a shared-memory configuration. The dual DSP controller permits the rangefinder robot to continuously perform several simultaneous tasks in real time, including the control of its motion and of its sensor platform, and the carrying out of Kalman filter calculations to estimate 18 state variables every 50 ms. Consequently, replication and adaptation of the hardware is now straightforward for further work and development in this area.
机译：协作导航系统（CNS）算法基于卡尔曼滤波器，该滤波器使用机器人之间的相对位移测量来改善一组自动移动机器人的集体位置估计。主要动机是当死角复航用作单个机器人的唯一导航方法时，减少位置误差的无限增长。本文建立了基于动力学的两驱动轮机器人组模型，并将其用于广泛的仿真中，以验证CNS算法的导航改进特性。这种机器人组的可观察性也得到了证明，并被用于预测导航误差仍然有界的情况以及误差是无界的情况。对于由编队运动的移动机器人组成的CNS组，发现了减少误差的下限的简单定量关系。 （多个机器人的）仿真和实验（使用一对小型的两个驱动轮机器人）表明，即使在两次机器人之间的测量之间存在延迟，也可以在正确的条件下接近该下限。论文中还包含了对两个实验开发必不可少的硬件开发项目的描述。首先是为测距仪机器人创建了一个仪器平台，使它能够获得携带信标的参考机器人的正确范围和方位，而忽略环境中的其他物体。第二项硬件开发任务是为测距仪机器人创建一个控制器，该控制器由两个德州仪器（TI）数字信号处理器（DSP）组成：共享内存配置中的TI TMS320F243和TMS320F2812。双DSP控制器使测距仪机器人能够连续实时地连续执行多个任务，包括对其运动和传感器平台的控制，以及进行卡尔曼滤波器计算以每50 ms估算18个状态变量。因此，硬件的复制和适配现在对于在该领域的进一步工作和开发而言是直接的。

著录项

  • 作者

    Fauconier, Richard.;

  • 作者单位

    Rensselaer Polytechnic Institute.;

  • 授予单位 Rensselaer Polytechnic Institute.;
  • 学科 Engineering Electronics and Electrical.; Artificial Intelligence.
  • 学位 Ph.D.
  • 年度 2004
  • 页码 166 p.
  • 总页数 166
  • 原文格式 PDF
  • 正文语种 eng
  • 中图分类 无线电电子学、电信技术;人工智能理论;
  • 关键词

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号